RepeatAfterMe

A package for the extension of repetitive DNA cores

The RepeatAfterMe RAMExtend tool automatically extends a multiple sequence alignment (MSA) that may represent only a fragment of a longer repetitive sequence family. Provided that the MSA includes detailed information on the coordinates for each sequence, RAMExtend will perform a local alignment extension of the MSA using flanking sequence. The tool generates consensus sequences for both the left and right extensions and can optionally output the full set of extended sequences in FASTA format.

The extension algorithm is an enhanced version of the RepeatScout approach developed by Alkes Price, Neil Jones and Pavel Pevzner (See History below). Enhancements include:

• Support for multiple scoring schemes
• Affine gap penalties for improved sensitivity
• Satellite detection to prevent extending repeats beyond a single unit

Author: Robert Hubley (2022–2025)
Institution: Institute for Systems Biology

RAMExtend

Overview

RAMExtend requires two input files:

1. Core alignment ranges (in a modified BED-6 format)
2. Genome sequence (in UCSC .2bit format)

The range file defines the alignment region for each sequence and which ends are eligible for extension.

Modified BED-6 Format

BED Field	Use by RAMExtend
chrom	Sequence identifier
chromStart	Lower aligned position (0-based)
chromEnd	Upper aligned position (half-open)
name	Left extendable flag (0 = no, 1 = yes)
score	Right extendable flag (0 = no, 1 = yes)
strand	'+' for forward, '-' for reverse

All fields are tab-separated and use 0-based, half-open coordinate conventions. The extendable flags control whether individual sequences participate in left/right extension—this is especially useful for sequence fragments in the core alignment which are not proximal to the core alignment edges.

Genome File

The genome must be in UCSC 2bit format, and should contain the sequences referenced by the BED file.

---

Example Usage

./RAMExtend -ranges test/extension-test2.tsv -twobit test/extension-test2.2bit

To output results to files:

./RAMExtend -ranges test/extension-test2.tsv -twobit test/extension-test2.2bit \
            -cons consensuses.fasta \
            -outtsv ext_ranges.tsv \
            -outfa ext_sequences.fasta

If no options are provided, usage help will be displayed.

---

extend-stk.pl

A wrapper script for RAMExtend that automatically extends and refines RepeatModeler seed alignments in Stockholm format. This helper script translates the Stockholm file into a TSV file for RAMExtend, invokes the tool, combines the left/right consensuses with the core alignment consensus, and rebuilds a new multiple sequence alignment using the core and extended sequences.

Example Usage

./util/extend-stk.pl -assembly test/ce10.2bit \
                     -input test/ce10-fam1.stk \
                     -output ce10-fam1-extended.stk

Example log:

##
## extend-stk.pl
##
##   Program Version      : 0.2
##   RAMExtend Version    : 0.0.6
##   RepeatModeler Version: 2.0.6
##   Genome               : test/ce10.2bit
##   Input                : test/ce10-fam1.stk
##   Output               : ce10-fam1-extended.stk
##   Min Aligning Seqs    : 3
##
Working on rnd-1_family-45...
  - Temporary directory: /u3/home/rhubley/projects/RepeatAfterMe-public/FDLO9azqC5
  - Consensus length [recalculated]: 163
  - Kimura divergence: 15.16% (no CpG adjustment)
  - Instances: 100
  - Running RAMExtend [bandwidth=40, matrix=14p43g, minimprovement=30]...
    - Estimated extensions: left 354 bp, right 86 bp, total 440
  - Rebuilding MSA with extensions...
    - Final consensus length = 603 [440 bp change]

---

Installation

Requirements

To build RepeatAfterMe, you will need:

• A C compiler that supports GNU11 (e.g. gcc)
• make
• POSIX-compatible environment (e.g., Linux/macOS)

Building

To compile the RAMExtend binary and the test suite:

make

This will:

• Compile the RAMExtend executable
• Build kentsrc/libTwoBit.a from source (included)
• Build a test_suite for validation and regression testing

Running Tests

Run the internal test suite:

./test_suite

Installing

Install the compiled binary and README to a system-wide or user-defined directory:

make install

By default, this installs into:

/usr/local/RepeatAfterMe-0.0.7/

You can add the binary to your PATH by:

export PATH=/usr/local/RepeatAfterMe-0.0.7/bin:$PATH

To change the install location, edit the INSTDIR variable near the top of the Makefile.

Cleaning the Build

To remove compiled artifacts:

make clean

---

History

RepeatAfterMe traces its roots to the pioneering work of Alkes Price, Neil Jones, and Pavel Pevzner, who developed an automated method for detecting repetitive DNA by building and extending multiple sequence alignments based on abundant exact words (k-mers). Their method was introduced in:

Price A.L., Jones N.C., Pevzner P.A. (2005).
De novo identification of repeat families in large genomes.
Proceedings of the 13th Annual International Conference on Intelligent Systems for Molecular Biology (ISMB-05), Detroit, Michigan.

One limitation of the original RepeatScout was its use of a simple match/mismatch/gap scoring system. This project began as an effort to add affine gap penalties and custom scoring matrices to RepeatScout, but evolved into a standalone tool when the scope of improvements grew.

RepeatAfterMe is intended as an experimental platform for applying this enhanced extension algorithm to various core types (e.g., k-mers, partial alignments). RAMExtend and extend-stk.pl are the first tools to leverage these algorithms.